An April 12 article in the Maritime Executive reports on a Rolls Royce statement that robot ships will be trading by 2020.
Apparently, through the Rolls-Royce led Advanced Autonomous Waterborne Applications Initiative (AAWA), researchers suggested that engineering hurdles would not be a major obstacle. There is no doubt that if engineering hurdles refer to hardware, that autonomous ships should be easier than autonomous aircraft or cars. A ship floats, and all it needs is a reliable means of propulsion to get where it needs to go using an autopilot and GPS.
In many ways the technology is already installed in ships and can be modified from off the shelf equipment even with inexpensive equipment used in recreational vessels.
If one wants to add rules of the road and accident avoidance one can add ARPA style radar and some programming, and maybe IR or LIDAR and to a large extent things are quite doable. But are these the actual engineering hurdles?
On a total ship system level, the engineering becomes much fuzzier, and upon closer examination I would argue that a robotic ship system is much more complicated than a robotic airplane or car system.
I really think true automated driving is around the corner, and automated flying is a given. We only keep pilots in the cockpit as a passenger assurance, which is sort of weird since they are excellent hijack targets and human error still counts for the majority of flying accidents.
So why am I down on autonomous ships? First let’s define autonomous ships. I am talking about a substantial size cargo vessel that carries cargo from country to country, and from random dock to random dock and that will autonomously operate from dock to dock simply with a destination command.
I know it can be done, but can it be done soon? To make it work means that the autonomous vehicle will be able to deal with common voyage occurrences such as:
- Loading: Some loading can be autonomous, but to make it happen would require a shore gang that consists of a berth mooring/loading party and a ship mooring/loading party. The ship party may be a locally hired gang, but that would require that the gang is intimately familiar with the ship. That requires a standardized design or an unusually expensive training load.
Undocking: In theory it is possible to make a ship unmoor autonomously, but such a system will be much more complex than straight mooring lines and will require shore compatibility. The question is: would it be economically viable to fit ships and shore with such systems. To date, except for ferries, there has been little work on such systems. Generally that means it is economically unattractive.
Outbound Pilotage: Pilotage is very heavily dependent on verbal (radio) communications. That contrasts with car operations where communications are visual and airplane operations where routes are very strict.
Equipment operation: No matter what anybody thinks, large ship operations today are far from turnkey. Simply dealing with heavy fuels is a complex process that requires decision making during the voyage and may result in system crashes, and has defied full automation. In theory, it would be possible to operate a vessel on distilled fuels for an entire voyage, but once purifiers and heavy fuel transfers come into play, things become much more difficult.
Equipment failure: Ships are subject to leaks. A leak is easy to detect (this is why ships have bilge alarms) but often the source of the leak is difficult to identify even with human means. As such, what may be a simple matter on a manned ship may quickly become a voyage failure on an unmanned ship.
Autonomy duration: Cars and airplanes do not operate on voyages that extend much beyond 12 hours, while ships may have to operate for days, if not weeks, without human intervention. A car can pull over to the side of the road and an airplane can land at the closest airport within three hours, but a ship may be days away from service, and in that time an initially harmless situation can have developed into a voyage failure.
Voyage threats: Ships do not operate in a friendly environment, they actually operate in the least friendly environment compared to airplanes and cars. Once things go wrong they can rapidly develop into disasters such as can occur with propulsion failures near reefs or rocky shores, near pirates, and in storms.
Inbound pilotage: Port operations may require that a vessel holds offshore or anchors for indeterminate times. Steaming in a holding pattern costs money and results in additional CO2 emissions. Unmanned anchoring is a fascinating process that can be accomplished with a heavily thrustered vessel, but thrusters are expensive and require maintenance. In theory it would be possible to fit an autonomous ship into a VTS, but in practice how many VTS systems are there? Airplanes have quite highly developed VTS’s, and so do cars (traffic lights) but such systems are much rarer for ships. For docking purposes, somehow lines need to be presented to the berth, does that mean that a gang will board the vessel when it comes along the dock? This is not a problem when there is high standardization, but can we count on high standardization?
Unloading: Unloading always requires a shore gang, but it also requires a surprisingly large amount of involvement from the ship’s point of view. Could these activities be automated or are they performed by a shore gang?
Equipment maintenance: This is the most interesting issue that appears to be unaddressed in all autonomous ship discussions. Today, only about three people of the vessel’s crew are involved in the navigation of ships. It is entirely possible that those three positions are eliminated with autonomous navigation. But as far as the portion of the crew that is involved in vessel maintenance is concerned, our efforts at reducing crews have stalled. On ocean going vessels, this part of the crew is rarely less than six people. If this maintenance personnel is not economically viable, we would have removed them from ships long ago, but somehow they keep hanging on. There is a simple reason for this; these people perform maintenance during the voyage that would otherwise have to occur ashore and shore based maintenance requires that a ship is tied up and a ship that is tied up loses money. As such, it is possible that a ship may be navigating autonomously, but it is unlikely it would be operating autonomously.
While I do not see the advent of fully autonomous ships as an economically viable proposition for deep ocean transportation, it could well be possible that we will start to see autonomous operation for certain short haul services and I could foresee autonomous docking tugs or other short voyage vessels (that inherently have long berth periods) in the near future. To a large extent we can autonomize cars and airplanes because there is a high level of uniformity with airplanes and cars (limited models and a well defined operating pattern). In maritime we achieved a huge efficiency increase through the uniformity of containerization, but it is much less clear if we can achieve the same efficiency gains if we only produce one size tanker with a standardized mooring, cargo and propulsion plant and one size bulker with a standardized mooring, cargo and propulsion plant, etc. etc. In other words, while the Rolls Royce statement may be true, it would only be true for a small portion of the marine adventure. So will we see any remote controlled ships in operation by the end of the decade? Yeah, probably. Will autonomous ships revolutionize the industry the way containerization changed the world? Probably not until after you and I will no longer need a driver’s license. Autonomous driving will be the real revolution; not autonomous navigation.